Let's say I have InfluxDB (2.7) measurements with two tags (id, site) and arbitrary fields. The id is fixed, and when the device is deployed to a different location, the site tag would change, which means that the old combination of (id, site=old_site) will stop receiving any more data (query is still possible). So, each site tag would have non-overlapping segments of time-series.
If I understand correctly, each unique (id, site) combination causes increased cardinality. I wonder if there is optimization in influxDB to "ignore" thoses inactive series? Would my abovementioned high-cardinality issue be less or equally concerning as the normal high-cardinality issue (many active tag combinations)?